Skip to content

models

Classes

fastvideo.configs.models.DiTConfig dataclass 

DiTConfig(arch_config: DiTArchConfig = DiTArchConfig(), prefix: str = '', quant_config: QuantizationConfig | None = None)

Bases: ModelConfig

Functions

fastvideo.configs.models.DiTConfig.add_cli_args staticmethod 
add_cli_args(parser: Any, prefix: str = 'dit-config') -> Any

Add CLI arguments for DiTConfig fields

Source code in fastvideo/configs/models/dits/base.py 
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
@staticmethod
def add_cli_args(parser: Any, prefix: str = "dit-config") -> Any:
    """Add CLI arguments for DiTConfig fields"""
    parser.add_argument(
        f"--{prefix}.prefix",
        type=str,
        dest=f"{prefix.replace('-', '_')}.prefix",
        default=DiTConfig.prefix,
        help="Prefix for the DiT model",
    )

    parser.add_argument(
        f"--{prefix}.quant-config",
        type=str,
        dest=f"{prefix.replace('-', '_')}.quant_config",
        default=None,
        help="Quantization configuration for the DiT model",
    )

    return parser

fastvideo.configs.models.VAEConfig dataclass 

VAEConfig(arch_config: VAEArchConfig = VAEArchConfig(), load_encoder: bool = True, load_decoder: bool = True, tile_sample_min_height: int = 256, tile_sample_min_width: int = 256, tile_sample_min_num_frames: int = 16, tile_sample_stride_height: int = 192, tile_sample_stride_width: int = 192, tile_sample_stride_num_frames: int = 12, blend_num_frames: int = 0, use_tiling: bool = True, use_temporal_tiling: bool = True, use_parallel_tiling: bool = True, use_temporal_scaling_frames: bool = True)

Bases: ModelConfig

Functions

fastvideo.configs.models.VAEConfig.add_cli_args staticmethod 
add_cli_args(parser: Any, prefix: str = 'vae-config') -> Any

Add CLI arguments for VAEConfig fields

Source code in fastvideo/configs/models/vaes/base.py 
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
@staticmethod
def add_cli_args(parser: Any, prefix: str = "vae-config") -> Any:
    """Add CLI arguments for VAEConfig fields"""
    parser.add_argument(
        f"--{prefix}.load-encoder",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.load_encoder",
        default=VAEConfig.load_encoder,
        help="Whether to load the VAE encoder",
    )
    parser.add_argument(
        f"--{prefix}.load-decoder",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.load_decoder",
        default=VAEConfig.load_decoder,
        help="Whether to load the VAE decoder",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-min-height",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_min_height",
        default=VAEConfig.tile_sample_min_height,
        help="Minimum height for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-min-width",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_min_width",
        default=VAEConfig.tile_sample_min_width,
        help="Minimum width for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-min-num-frames",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_min_num_frames",
        default=VAEConfig.tile_sample_min_num_frames,
        help="Minimum number of frames for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-stride-height",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_stride_height",
        default=VAEConfig.tile_sample_stride_height,
        help="Stride height for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-stride-width",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_stride_width",
        default=VAEConfig.tile_sample_stride_width,
        help="Stride width for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-stride-num-frames",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_stride_num_frames",
        default=VAEConfig.tile_sample_stride_num_frames,
        help="Stride number of frames for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.blend-num-frames",
        type=int,
        dest=f"{prefix.replace('-', '_')}.blend_num_frames",
        default=VAEConfig.blend_num_frames,
        help="Number of frames to blend for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.use-tiling",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.use_tiling",
        default=VAEConfig.use_tiling,
        help="Whether to use tiling for VAE",
    )
    parser.add_argument(
        f"--{prefix}.use-temporal-tiling",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.use_temporal_tiling",
        default=VAEConfig.use_temporal_tiling,
        help="Whether to use temporal tiling for VAE",
    )
    parser.add_argument(
        f"--{prefix}.use-parallel-tiling",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.use_parallel_tiling",
        default=VAEConfig.use_parallel_tiling,
        help="Whether to use parallel tiling for VAE",
    )

    return parser

Modules

fastvideo.configs.models.dits

Classes

fastvideo.configs.models.dits.Cosmos25VideoConfig dataclass 
Cosmos25VideoConfig(arch_config: DiTArchConfig = Cosmos25ArchConfig(), prefix: str = 'Cosmos25', quant_config: QuantizationConfig | None = None)

Bases: DiTConfig

Configuration for Cosmos 2.5 video generation model.

Modules

fastvideo.configs.models.dits.base
Classes
fastvideo.configs.models.dits.base.DiTConfig dataclass 
DiTConfig(arch_config: DiTArchConfig = DiTArchConfig(), prefix: str = '', quant_config: QuantizationConfig | None = None)

Bases: ModelConfig

Functions
fastvideo.configs.models.dits.base.DiTConfig.add_cli_args staticmethod 
add_cli_args(parser: Any, prefix: str = 'dit-config') -> Any

Add CLI arguments for DiTConfig fields

Source code in fastvideo/configs/models/dits/base.py 
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
@staticmethod
def add_cli_args(parser: Any, prefix: str = "dit-config") -> Any:
    """Add CLI arguments for DiTConfig fields"""
    parser.add_argument(
        f"--{prefix}.prefix",
        type=str,
        dest=f"{prefix.replace('-', '_')}.prefix",
        default=DiTConfig.prefix,
        help="Prefix for the DiT model",
    )

    parser.add_argument(
        f"--{prefix}.quant-config",
        type=str,
        dest=f"{prefix.replace('-', '_')}.quant_config",
        default=None,
        help="Quantization configuration for the DiT model",
    )

    return parser
fastvideo.configs.models.dits.cosmos2_5
Classes
fastvideo.configs.models.dits.cosmos2_5.Cosmos25ArchConfig dataclass 
Cosmos25ArchConfig(stacked_params_mapping: list[tuple[str, str, str]] = list(), _fsdp_shard_conditions: list = (lambda: [is_transformer_blocks])(), _compile_conditions: list = list(), param_names_mapping: dict = (lambda: {'^net\\.x_embedder\\.proj\\.1\\.(.*)$': 'patch_embed.proj.\\1', '^net\\.t_embedder\\.1\\.linear_1\\.(.*)$': 'time_embed.t_embedder.linear_1.\\1', '^net\\.t_embedder\\.1\\.linear_2\\.(.*)$': 'time_embed.t_embedder.linear_2.\\1', '^net\\.t_embedding_norm\\.(.*)$': 'time_embed.norm.\\1', '^net\\.crossattn_proj\\.0\\.weight$': 'crossattn_proj.0.weight', '^net\\.crossattn_proj\\.0\\.bias$': 'crossattn_proj.0.bias', '^net\\.blocks\\.(\\d+)\\.self_attn\\.q_proj\\.(.*)$': 'transformer_blocks.\\1.attn1.to_q.\\2', '^net\\.blocks\\.(\\d+)\\.self_attn\\.k_proj\\.(.*)$': 'transformer_blocks.\\1.attn1.to_k.\\2', '^net\\.blocks\\.(\\d+)\\.self_attn\\.v_proj\\.(.*)$': 'transformer_blocks.\\1.attn1.to_v.\\2', '^net\\.blocks\\.(\\d+)\\.self_attn\\.output_proj\\.(.*)$': 'transformer_blocks.\\1.attn1.to_out.\\2', '^net\\.blocks\\.(\\d+)\\.self_attn\\.q_norm\\.weight$': 'transformer_blocks.\\1.attn1.norm_q.weight', '^net\\.blocks\\.(\\d+)\\.self_attn\\.k_norm\\.weight$': 'transformer_blocks.\\1.attn1.norm_k.weight', '^net\\.blocks\\.(\\d+)\\.self_attn\\.q_norm\\._extra_state$': 'transformer_blocks.\\1.attn1.norm_q._extra_state', '^net\\.blocks\\.(\\d+)\\.self_attn\\.k_norm\\._extra_state$': 'transformer_blocks.\\1.attn1.norm_k._extra_state', '^net\\.blocks\\.(\\d+)\\.cross_attn\\.q_proj\\.(.*)$': 'transformer_blocks.\\1.attn2.to_q.\\2', '^net\\.blocks\\.(\\d+)\\.cross_attn\\.k_proj\\.(.*)$': 'transformer_blocks.\\1.attn2.to_k.\\2', '^net\\.blocks\\.(\\d+)\\.cross_attn\\.v_proj\\.(.*)$': 'transformer_blocks.\\1.attn2.to_v.\\2', '^net\\.blocks\\.(\\d+)\\.cross_attn\\.output_proj\\.(.*)$': 'transformer_blocks.\\1.attn2.to_out.\\2', '^net\\.blocks\\.(\\d+)\\.cross_attn\\.q_norm\\.weight$': 'transformer_blocks.\\1.attn2.norm_q.weight', '^net\\.blocks\\.(\\d+)\\.cross_attn\\.k_norm\\.weight$': 'transformer_blocks.\\1.attn2.norm_k.weight', '^net\\.blocks\\.(\\d+)\\.cross_attn\\.q_norm\\._extra_state$': 'transformer_blocks.\\1.attn2.norm_q._extra_state', '^net\\.blocks\\.(\\d+)\\.cross_attn\\.k_norm\\._extra_state$': 'transformer_blocks.\\1.attn2.norm_k._extra_state', '^net\\.blocks\\.(\\d+)\\.mlp\\.layer1\\.(.*)$': 'transformer_blocks.\\1.mlp.fc_in.\\2', '^net\\.blocks\\.(\\d+)\\.mlp\\.layer2\\.(.*)$': 'transformer_blocks.\\1.mlp.fc_out.\\2', '^net\\.blocks\\.(\\d+)\\.adaln_modulation_self_attn\\.1\\.(.*)$': 'transformer_blocks.\\1.adaln_modulation_self_attn.1.\\2', '^net\\.blocks\\.(\\d+)\\.adaln_modulation_self_attn\\.2\\.(.*)$': 'transformer_blocks.\\1.adaln_modulation_self_attn.2.\\2', '^net\\.blocks\\.(\\d+)\\.adaln_modulation_cross_attn\\.1\\.(.*)$': 'transformer_blocks.\\1.adaln_modulation_cross_attn.1.\\2', '^net\\.blocks\\.(\\d+)\\.adaln_modulation_cross_attn\\.2\\.(.*)$': 'transformer_blocks.\\1.adaln_modulation_cross_attn.2.\\2', '^net\\.blocks\\.(\\d+)\\.adaln_modulation_mlp\\.1\\.(.*)$': 'transformer_blocks.\\1.adaln_modulation_mlp.1.\\2', '^net\\.blocks\\.(\\d+)\\.adaln_modulation_mlp\\.2\\.(.*)$': 'transformer_blocks.\\1.adaln_modulation_mlp.2.\\2', '^net\\.blocks\\.(\\d+)\\.layer_norm_self_attn\\._extra_state$': 'transformer_blocks.\\1.norm1.norm._extra_state', '^net\\.blocks\\.(\\d+)\\.layer_norm_cross_attn\\._extra_state$': 'transformer_blocks.\\1.norm2.norm._extra_state', '^net\\.blocks\\.(\\d+)\\.layer_norm_mlp\\._extra_state$': 'transformer_blocks.\\1.norm3.norm._extra_state', '^net\\.final_layer\\.linear\\.(.*)$': 'final_layer.proj_out.\\1', '^net\\.final_layer\\.adaln_modulation\\.1\\.(.*)$': 'final_layer.linear_1.\\1', '^net\\.final_layer\\.adaln_modulation\\.2\\.(.*)$': 'final_layer.linear_2.\\1'})(), reverse_param_names_mapping: dict = dict(), lora_param_names_mapping: dict = (lambda: {'^transformer_blocks\\.(\\d+)\\.attn1\\.to_q\\.(.*)$': 'transformer_blocks.\\1.attn1.to_q.\\2', '^transformer_blocks\\.(\\d+)\\.attn1\\.to_k\\.(.*)$': 'transformer_blocks.\\1.attn1.to_k.\\2', '^transformer_blocks\\.(\\d+)\\.attn1\\.to_v\\.(.*)$': 'transformer_blocks.\\1.attn1.to_v.\\2', '^transformer_blocks\\.(\\d+)\\.attn1\\.to_out\\.(.*)$': 'transformer_blocks.\\1.attn1.to_out.\\2', '^transformer_blocks\\.(\\d+)\\.attn2\\.to_q\\.(.*)$': 'transformer_blocks.\\1.attn2.to_q.\\2', '^transformer_blocks\\.(\\d+)\\.attn2\\.to_k\\.(.*)$': 'transformer_blocks.\\1.attn2.to_k.\\2', '^transformer_blocks\\.(\\d+)\\.attn2\\.to_v\\.(.*)$': 'transformer_blocks.\\1.attn2.to_v.\\2', '^transformer_blocks\\.(\\d+)\\.attn2\\.to_out\\.(.*)$': 'transformer_blocks.\\1.attn2.to_out.\\2', '^transformer_blocks\\.(\\d+)\\.mlp\\.(.*)$': 'transformer_blocks.\\1.mlp.\\2'})(), _supported_attention_backends: tuple[AttentionBackendEnum, ...] = (SLIDING_TILE_ATTN, SAGE_ATTN, FLASH_ATTN, TORCH_SDPA, VIDEO_SPARSE_ATTN, VMOBA_ATTN, SAGE_ATTN_THREE), hidden_size: int = 0, num_attention_heads: int = 16, num_channels_latents: int = 0, exclude_lora_layers: list[str] = (lambda: ['embedder'])(), boundary_ratio: float | None = None, in_channels: int = 16, out_channels: int = 16, attention_head_dim: int = 128, num_layers: int = 28, mlp_ratio: float = 4.0, text_embed_dim: int = 1024, adaln_lora_dim: int = 256, use_adaln_lora: bool = True, max_size: tuple[int, int, int] = (128, 240, 240), patch_size: tuple[int, int, int] = (1, 2, 2), rope_scale: tuple[float, float, float] = (1.0, 3.0, 3.0), concat_padding_mask: bool = True, extra_pos_embed_type: str | None = None, use_crossattn_projection: bool = False, crossattn_proj_in_channels: int = 100352, rope_enable_fps_modulation: bool = True, qk_norm: str = 'rms_norm', eps: float = 1e-06)

Bases: DiTArchConfig

Configuration for Cosmos 2.5 architecture (MiniTrainDIT).

fastvideo.configs.models.dits.cosmos2_5.Cosmos25VideoConfig dataclass 
Cosmos25VideoConfig(arch_config: DiTArchConfig = Cosmos25ArchConfig(), prefix: str = 'Cosmos25', quant_config: QuantizationConfig | None = None)

Bases: DiTConfig

Configuration for Cosmos 2.5 video generation model.

fastvideo.configs.models.encoders

Classes

fastvideo.configs.models.encoders.T5LargeConfig dataclass 
T5LargeConfig(arch_config: TextEncoderArchConfig = T5LargeArchConfig(), prefix: str = 't5', quant_config: QuantizationConfig | None = None, lora_config: Any | None = None)

Bases: TextEncoderConfig

T5 Large configuration for your specific model.

Modules

fastvideo.configs.models.encoders.t5
Classes
fastvideo.configs.models.encoders.t5.T5LargeArchConfig dataclass 
T5LargeArchConfig(stacked_params_mapping: list[tuple[str, str, str]] = (lambda: [('.qkv_proj', '.q', 'q'), ('.qkv_proj', '.k', 'k'), ('.qkv_proj', '.v', 'v')])(), architectures: list[str] = (lambda: [])(), _supported_attention_backends: tuple[AttentionBackendEnum, ...] = (FLASH_ATTN, TORCH_SDPA), output_hidden_states: bool = False, use_return_dict: bool = True, vocab_size: int = 32128, hidden_size: int = 0, num_hidden_layers: int = 0, num_attention_heads: int = 0, pad_token_id: int = 0, eos_token_id: int = 1, text_len: int = 512, hidden_state_skip_layer: int = 0, decoder_start_token_id: int = 0, output_past: bool = True, scalable_attention: bool = True, tie_word_embeddings: bool = False, tokenizer_kwargs: dict[str, Any] = dict(), _fsdp_shard_conditions: list = (lambda: [_is_transformer_layer, _is_embeddings, _is_final_layernorm])(), d_model: int = 1024, d_kv: int = 128, d_ff: int = 65536, num_layers: int = 24, num_decoder_layers: int | None = 24, num_heads: int = 128, relative_attention_num_buckets: int = 32, relative_attention_max_distance: int = 128, dropout_rate: float = 0.1, layer_norm_epsilon: float = 1e-06, initializer_factor: float = 1.0, feed_forward_proj: str = 'relu', dense_act_fn: str = '', is_gated_act: bool = False, is_encoder_decoder: bool = True, use_cache: bool = True, classifier_dropout: float = 0.0, n_positions: int = 512, task_specific_params: dict | None = None)

Bases: T5ArchConfig

T5 Large architecture config with parameters for your specific model.

fastvideo.configs.models.encoders.t5.T5LargeConfig dataclass 
T5LargeConfig(arch_config: TextEncoderArchConfig = T5LargeArchConfig(), prefix: str = 't5', quant_config: QuantizationConfig | None = None, lora_config: Any | None = None)

Bases: TextEncoderConfig

T5 Large configuration for your specific model.

fastvideo.configs.models.vaes

Modules

fastvideo.configs.models.vaes.base
Classes
fastvideo.configs.models.vaes.base.VAEConfig dataclass 
VAEConfig(arch_config: VAEArchConfig = VAEArchConfig(), load_encoder: bool = True, load_decoder: bool = True, tile_sample_min_height: int = 256, tile_sample_min_width: int = 256, tile_sample_min_num_frames: int = 16, tile_sample_stride_height: int = 192, tile_sample_stride_width: int = 192, tile_sample_stride_num_frames: int = 12, blend_num_frames: int = 0, use_tiling: bool = True, use_temporal_tiling: bool = True, use_parallel_tiling: bool = True, use_temporal_scaling_frames: bool = True)

Bases: ModelConfig

Functions
fastvideo.configs.models.vaes.base.VAEConfig.add_cli_args staticmethod 
add_cli_args(parser: Any, prefix: str = 'vae-config') -> Any

Add CLI arguments for VAEConfig fields

Source code in fastvideo/configs/models/vaes/base.py 
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
@staticmethod
def add_cli_args(parser: Any, prefix: str = "vae-config") -> Any:
    """Add CLI arguments for VAEConfig fields"""
    parser.add_argument(
        f"--{prefix}.load-encoder",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.load_encoder",
        default=VAEConfig.load_encoder,
        help="Whether to load the VAE encoder",
    )
    parser.add_argument(
        f"--{prefix}.load-decoder",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.load_decoder",
        default=VAEConfig.load_decoder,
        help="Whether to load the VAE decoder",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-min-height",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_min_height",
        default=VAEConfig.tile_sample_min_height,
        help="Minimum height for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-min-width",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_min_width",
        default=VAEConfig.tile_sample_min_width,
        help="Minimum width for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-min-num-frames",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_min_num_frames",
        default=VAEConfig.tile_sample_min_num_frames,
        help="Minimum number of frames for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-stride-height",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_stride_height",
        default=VAEConfig.tile_sample_stride_height,
        help="Stride height for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-stride-width",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_stride_width",
        default=VAEConfig.tile_sample_stride_width,
        help="Stride width for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.tile-sample-stride-num-frames",
        type=int,
        dest=f"{prefix.replace('-', '_')}.tile_sample_stride_num_frames",
        default=VAEConfig.tile_sample_stride_num_frames,
        help="Stride number of frames for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.blend-num-frames",
        type=int,
        dest=f"{prefix.replace('-', '_')}.blend_num_frames",
        default=VAEConfig.blend_num_frames,
        help="Number of frames to blend for VAE tile sampling",
    )
    parser.add_argument(
        f"--{prefix}.use-tiling",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.use_tiling",
        default=VAEConfig.use_tiling,
        help="Whether to use tiling for VAE",
    )
    parser.add_argument(
        f"--{prefix}.use-temporal-tiling",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.use_temporal_tiling",
        default=VAEConfig.use_temporal_tiling,
        help="Whether to use temporal tiling for VAE",
    )
    parser.add_argument(
        f"--{prefix}.use-parallel-tiling",
        action=StoreBoolean,
        dest=f"{prefix.replace('-', '_')}.use_parallel_tiling",
        default=VAEConfig.use_parallel_tiling,
        help="Whether to use parallel tiling for VAE",
    )

    return parser